Deep Learning for Motor Imagery Brain-Computer Interfaces
Systematic review of MI EEG deep learning work (2015–2024), tracing the shift from CNNs to Transformers.
Active
PythonDeep LearningPyTorchEEG
Systematic Review: Deep Learning for Motor Imagery
Brain-Computer Interfaces survey tracing the shift from CNNs to Transformers.
Key Findings
- Traced the shift from CNNs to Transformers and evaluated reproducibility (26.4%).
- Model scales (2.6K–12.8M) and compute needs.
- Demonstrated 20–25% improvements in cross-subject generalization with TL methods.